7
High-Resolution Double-Quantum 31 P MAS NMR Study of the Intermediate-Range Order in Crystalline and Glass Lead Phosphates Franck Fayon,* Catherine Bessada, Jean-Pierre Coutures, and Dominique Massiot Centre de Recherche sur les Mate ´riaux a ` Hautes Tempe ´ratures, CNRS, 1D Avenue de la Recherche Scientifique, 45071 Orle ´ans Cedex 2, France ReceiVed April 6, 1999 We have investigated the structure of binary lead phosphate glasses using 1D 31 P MAS and 2D 31 P double- quantum NMR. The distribution of the Q n phosphate species, determined from the simulation of 1D MAS spectra, indicates a significant disproportionation reaction (2Q n T Q n+1 + Q n-1 ) near the pyrophosphate composition. As demonstrated on crystalline reference samples, 2D double-quantum experiments can be used to determine the connectivities between Q n groups. This allows us to probe the phosphorus connectivity scheme in the glass network and to show that the Q n chemical shift is influenced by the type of the bonded PO 4 tetrahedron. By analyzing quantitative 1D MAS spectra and 2D double-quantum spectra, we have determined the phosphorus connectivity scheme in the glass structure and its evolution with composition. Introduction Recently, a whole range of new applications of phosphate- based glasses have been developed 1,2 (laser glasses, optoelec- tronics, nuclear waste storage). A clear characterization and understanding of their structure on different length scales is thus highly desirable but remains experimentally difficult, mainly due to their lack of periodicity. Different techniques such as Raman scattering, XPS, IR and NMR spectroscopies have already been used to study the local ordering around phospho- rus. 3 In phosphate glasses, phosphorus always occurs in tetrahedral coordination. The different tetrahedra can be clas- sified according to their connectivities, Q n , where n represents the number of bridging oxygen atoms per PO 4 tetrahedron (n ) 0-3). However, the description of the intermediate range order in these systems, i.e., the ordering of basic structural units within the second and third coordination spheres of the phosphorus atoms, still remains a question. Solid-state nuclear magnetic resonance (NMR) spectroscopy has proved to be a powerful tool for the structural analysis of disordered solids such as glasses. 4 In particular, the different Q n units constituting the phosphate network have been clearly evidenced in high- resolution magic angle spinning (MAS) NMR spectra through the 31 P chemical shift interaction which is sensitive to the phosphorus local environment. Furthermore, the through-space magnetic dipolar interaction strongly depends on the inter-nuclei distance and can be used to probe the spatial connectivity between atoms in solids. This interaction is averaged out under fast MAS spinning required for obtaining high-resolution spectra. In the case of spin 1 / 2 nuclei like 31 P, several pulse sequences have been developed to selectively reintroduce the homonuclear dipolar interaction in rotating solids. 5-11 This allows one to obtain a MAS-resolved spectrum that provides both the chemical shift and the homonuclear dipolar information. These sequences, designed for two-dimensional magnetization exchange experiments or double-quantum NMR spectroscopy, have been recently employed with success to characterize the phosphorus connectivity scheme in crystalline phosphates 12,13 and glasses. 14-18 In the case of the binary lead phosphate glasses, the local lead environment has previously been probed by EXAFS, 19 X-ray diffraction, 20 and 207 Pb NMR. 21 However, the structure of the phosphate network has not been thoroughly investigated. In this work, we used 31 P one-dimensional MAS and two- dimensional double-quantum NMR to study the short and intermediate range structure of crystalline phases and glasses in the binary xPbO(1-x)P 2 O 5 system with x varying from 0.5 to 0.67. (1) Weber, M. J. J. Non-Cryst. Solids 1990, 123, 208-222. (2) Sales, B. C.; Boatner, L. A. Science 1984, 226, 45-48. (3) Martin, S. W. Eur. J. Solid State Inorg. Chem. 1991, 28, 163-205. (4) Eckert, H. Prog. Nucl. Magn. Reson. Spectrosc. 1992, 24, 159-293. (5) Tycko, R.; Dabbagh, G. J. Am. Chem. Soc. 1991, 113, 9444-9448. (6) Bennett, A. E.; Ok, J. H.; Griffin, R. G.; Vega, S. J. Chem. Phys. 1992, 96, 8624-8627. (7) Nielsen, N. C.; Bildsøe, H.; Jakobsen, H. J.; Levitt, M. H. J. Chem. Phys. 1994, 101, 1806-1812. (8) Sun, B. Q.; Costa, P. R.; Kocisko, D.; Lansbury, P. T.; Griffin, R. G. J. Chem. Phys. 1995, 102, 702-707. (9) Baldus, M.; Tomaselli, M.; Meier, B. H.; Ernst, R. R. Chem. Phys. Lett. 1994, 230, 329-336. (10) Lee, Y. K.; Kurur, N. D.; Elm, M.; Johannessen, O. G.; Nielsen, N. C.; Levitt, M. H. Chem. Phys. Lett. 1995, 242, 304-309. (11) Feike, M.; Demco, D. E.; Graf, R.; Gottwald, J.; Hafner, S.; Spiess, H. W. J. Magn. Reson. A 1996, 122, 214-221. (12) Feike, M.; Graf, R.; Schnell, I.; Ja ¨ger, C.; Spiess, H. W. J. Am. Chem. Soc. 1996, 118, 9631-9634. (13) Geen, H.; Gottwald, J.; Graf, R.; Schnell, I.; Spiess, H. W.; Titman, J. J. J. Magn. Reson. 1997, 125, 224-227. (14) Ja ¨ger, C.; Feike, M.; Born, R.; Spiess, H. W. J. Non-Cryst. Solids 1994, 180, 91-95. (15) Olsen, K. K.; Zwanziger, J. W.; Hartmann, P.; Ja ¨ger, C. J. Non-Cryst. Solids 1997, 222, 199-205. (16) Alam, T. M.; Brow, R. K. J. Non-Cryst. Solids 1998, 223,1-20. (17) Feike, M.; Ja ¨ger, C.; Spiess, H. W. J. Non-Cryst. Solids 1998, 223, 200-206. (18) Witter, R.; Hartmann, P.; Vogel, J.; Ja ¨ger, C. Solid State NMR 1998, 13, 189-200. (19) Greaves, G. N.; Gurman, S. J.; Gladden, L. F.; Spence, C. A.; Cox, P.; Sales, B. C.; Boatner, L. A.; Jenkins, R. N. Philos. Mag. 1988, B58, 271-283. (20) Lai, A.; Musinu, A.; Piccaluga, G.; Puligheddu, S. Phys. Chem. Glasses 1997, 38, 173-178. (21) Fayon, F.; Bessada, C.; Douy, A.; Massiot, D. J. Magn. Reson. 1999, 137, 116-121. 5212 Inorg. Chem. 1999, 38, 5212-5218 10.1021/ic990375p CCC: $18.00 © 1999 American Chemical Society Published on Web 10/21/1999

High-Resolution Double-Quantum 31 P MAS NMR Study of the Intermediate-Range Order in Crystalline and Glass Lead Phosphates

Embed Size (px)

Citation preview

Page 1: High-Resolution Double-Quantum               31               P MAS NMR Study of the Intermediate-Range Order in Crystalline and Glass Lead Phosphates

High-Resolution Double-Quantum31P MAS NMR Study of the Intermediate-Range Orderin Crystalline and Glass Lead Phosphates

Franck Fayon,* Catherine Bessada, Jean-Pierre Coutures, and Dominique Massiot

Centre de Recherche sur les Mate´riaux aHautes Tempe´ratures, CNRS,1D Avenue de la Recherche Scientifique, 45071 Orle´ans Cedex 2, France

ReceiVed April 6, 1999

We have investigated the structure of binary lead phosphate glasses using 1D31P MAS and 2D31P double-quantum NMR. The distribution of the Qn phosphate species, determined from the simulation of 1D MAS spectra,indicates a significant disproportionation reaction (2Qn T Qn+1 + Qn-1) near the pyrophosphate composition. Asdemonstrated on crystalline reference samples, 2D double-quantum experiments can be used to determine theconnectivities between Qn groups. This allows us to probe the phosphorus connectivity scheme in the glass networkand to show that the Qn chemical shift is influenced by the type of the bonded PO4 tetrahedron. By analyzingquantitative 1D MAS spectra and 2D double-quantum spectra, we have determined the phosphorus connectivityscheme in the glass structure and its evolution with composition.

Introduction

Recently, a whole range of new applications of phosphate-based glasses have been developed1,2 (laser glasses, optoelec-tronics, nuclear waste storage). A clear characterization andunderstanding of their structure on different length scales is thushighly desirable but remains experimentally difficult, mainlydue to their lack of periodicity. Different techniques such asRaman scattering, XPS, IR and NMR spectroscopies havealready been used to study the local ordering around phospho-rus.3 In phosphate glasses, phosphorus always occurs intetrahedral coordination. The different tetrahedra can be clas-sified according to their connectivities, Qn, wheren representsthe number of bridging oxygen atoms per PO4 tetrahedron (n) 0-3). However, the description of the intermediate rangeorder in these systems, i.e., the ordering of basic structural unitswithin the second and third coordination spheres of thephosphorus atoms, still remains a question. Solid-state nuclearmagnetic resonance (NMR) spectroscopy has proved to be apowerful tool for the structural analysis of disordered solidssuch as glasses.4 In particular, the different Qn units constitutingthe phosphate network have been clearly evidenced in high-resolution magic angle spinning (MAS) NMR spectra throughthe 31P chemical shift interaction which is sensitive to thephosphorus local environment. Furthermore, the through-spacemagnetic dipolar interaction strongly depends on the inter-nucleidistance and can be used to probe the spatial connectivitybetween atoms in solids. This interaction is averaged out underfast MAS spinning required for obtaining high-resolutionspectra. In the case of spin1/2 nuclei like 31P, several pulsesequences have been developed to selectively reintroduce thehomonuclear dipolar interaction in rotating solids.5-11 Thisallows one to obtain a MAS-resolved spectrum that provides

both the chemical shift and the homonuclear dipolar information.These sequences, designed for two-dimensional magnetizationexchange experiments or double-quantum NMR spectroscopy,have been recently employed with success to characterize thephosphorus connectivity scheme in crystalline phosphates12,13

and glasses.14-18

In the case of the binary lead phosphate glasses, the locallead environment has previously been probed by EXAFS,19

X-ray diffraction,20 and 207Pb NMR.21 However, the structureof the phosphate network has not been thoroughly investigated.In this work, we used31P one-dimensional MAS and two-dimensional double-quantum NMR to study the short andintermediate range structure of crystalline phases and glassesin the binaryxPbO(1-x)P2O5 system withx varying from 0.5to 0.67.

(1) Weber, M. J.J. Non-Cryst. Solids1990, 123, 208-222.(2) Sales, B. C.; Boatner, L. A.Science1984, 226, 45-48.(3) Martin, S. W.Eur. J. Solid State Inorg. Chem.1991, 28, 163-205.(4) Eckert, H.Prog. Nucl. Magn. Reson. Spectrosc.1992, 24, 159-293.(5) Tycko, R.; Dabbagh, G.J. Am. Chem. Soc.1991, 113, 9444-9448.(6) Bennett, A. E.; Ok, J. H.; Griffin, R. G.; Vega, S.J. Chem. Phys.

1992, 96, 8624-8627.

(7) Nielsen, N. C.; Bildsøe, H.; Jakobsen, H. J.; Levitt, M. H.J. Chem.Phys.1994, 101, 1806-1812.

(8) Sun, B. Q.; Costa, P. R.; Kocisko, D.; Lansbury, P. T.; Griffin, R. G.J. Chem. Phys.1995, 102, 702-707.

(9) Baldus, M.; Tomaselli, M.; Meier, B. H.; Ernst, R. R.Chem. Phys.Lett. 1994, 230, 329-336.

(10) Lee, Y. K.; Kurur, N. D.; Elm, M.; Johannessen, O. G.; Nielsen, N.C.; Levitt, M. H. Chem. Phys. Lett.1995, 242, 304-309.

(11) Feike, M.; Demco, D. E.; Graf, R.; Gottwald, J.; Hafner, S.; Spiess,H. W. J. Magn. Reson. A1996, 122, 214-221.

(12) Feike, M.; Graf, R.; Schnell, I.; Ja¨ger, C.; Spiess, H. W.J. Am. Chem.Soc.1996, 118, 9631-9634.

(13) Geen, H.; Gottwald, J.; Graf, R.; Schnell, I.; Spiess, H. W.; Titman,J. J.J. Magn. Reson.1997, 125, 224-227.

(14) Jager, C.; Feike, M.; Born, R.; Spiess, H. W.J. Non-Cryst. Solids1994, 180, 91-95.

(15) Olsen, K. K.; Zwanziger, J. W.; Hartmann, P.; Ja¨ger, C.J. Non-Cryst.Solids1997, 222, 199-205.

(16) Alam, T. M.; Brow, R. K.J. Non-Cryst. Solids1998, 223, 1-20.(17) Feike, M.; Ja¨ger, C.; Spiess, H. W.J. Non-Cryst. Solids1998, 223,

200-206.(18) Witter, R.; Hartmann, P.; Vogel, J.; Ja¨ger, C.Solid State NMR1998,

13, 189-200.(19) Greaves, G. N.; Gurman, S. J.; Gladden, L. F.; Spence, C. A.; Cox,

P.; Sales, B. C.; Boatner, L. A.; Jenkins, R. N.Philos. Mag.1988,B58, 271-283.

(20) Lai, A.; Musinu, A.; Piccaluga, G.; Puligheddu, S.Phys. Chem. Glasses1997, 38, 173-178.

(21) Fayon, F.; Bessada, C.; Douy, A.; Massiot, D.J. Magn. Reson.1999,137, 116-121.

5212 Inorg. Chem.1999,38, 5212-5218

10.1021/ic990375p CCC: $18.00 © 1999 American Chemical SocietyPublished on Web 10/21/1999

Page 2: High-Resolution Double-Quantum               31               P MAS NMR Study of the Intermediate-Range Order in Crystalline and Glass Lead Phosphates

Experimental Section

The P2O5-PbO glasses were prepared from reagent grade PbO and(NH4)2HPO4. The glass samples were melted in Pt crucibles for 1h inan electric furnace under air at temperature varying from 700 to 850°C depending on composition and quenched by partly immersing thecrucible in water. An additional 0.02 mol % of Fe2O3 was included inthe batch to reduce the31P relaxation time. Glass compositions werechecked by microprobe (1 mol % uncertainties). The Pb2P2O7, Pb3P4O13,and Pb(PO3)2 crystalline phases were obtained by annealing the glassesat high temperature and checked by powder X-ray diffraction.

The solid-state NMR experiments were performed on a Bruker DSX300 spectrometer with a 4 mm MASprobehead operating at a Larmorfrequency of 121.4 MHz for31P. Theπ/2 pulse duration was 3.5µs(ωrf/2π ) 71.4 kHz). The one-dimensional (1D)31P MAS spectra wererecorded at 3 and 10 kHz spinning rates using single pulse (π/12)acquisition with recycle delay varying from 2 to 20 s to ensure nosaturation.

The two-dimensional (2D)31P double-quantum NMR spectra wereacquired while spinning at 10 kHz using the pulse sequence depictedin Figure 1. The aim of this experiment is to correlate the usual 1DMAS spectrum, characteristic of the individual31P resonances, to thedouble-quantum spectrum, characteristic of the resonance of pairs ofnearby31P atoms coupled by dipolar interaction. The initial excitationperiod is used to create the double-quantum coherence which is allowedto evolve during an incremented timet1. During the reconversion period,the double-quantum coherence is reconverted to Zeeman magnetizationand the finalπ/2 pulse is applied to create transverse magnetizationacquired during the detection timet2 (FID). The t1 modulated signalacquired duringt2 is then doubly Fourier transformed (according tot1and t2 times) to obtain a two-dimensional correlation spectrum thatcorrelates a double-quantum spectrum to a “double-quantum filtered”single-quantum MAS spectrum. A pair of coupled nuclei, whichresonate individually at frequenciesωA andωB in the single-quantumω2 dimension, resonate at the sum of the frequencies (ωA + ωB) in thedouble-quantumω1 dimension. Consequently, double-quantum coher-ence involving two distinct sites is evidenced by twin cross-correlationpeaks on each side of the diagonal of the two-dimensional spectrum,while double-quantum coherence between two equivalent sites gives asingle autocorrelation peak on the diagonal. Since the homonucleardipolar interaction is averaged out by magic angle spinning, a recouplingsequence has to be used to create a dipolar double-quantum Hamilto-nian. We have used the C7 recoupling sequence10 to achieve an efficientbroadband excitation and reconversion of double-quantum coherencesunder MAS conditions (ωrf/2π ) 69.5 kHz at a spinning rate of 10kHz). The excitation and reconversion periods were kept short (600µs), and under these conditions, the intensity of double-quantumresonances is proportional to the square of the dipolar coupling and tothe number of coupled spin pairs.22 The phases of the C7 excitation

sequence and of the lastπ/2 pulse were cycled by step of 90° to selectthe coherence pathway of Figure 1. The two-dimensional pure absorp-tion phase spectra were obtained using a hypercomplex acquisition.23

Processing of the 2D spectra was made using the RMN program.24 Toavoid spinning sidebands in theω1 dimension, thet1 time incrementwas synchronized with the rotor period; 128 and 64t1 increments weretaken for the crystalline and glass samples, respectively. Each slicewas the sum of 16 transients using a recycle delay of 20 s. The entireexperimental time required to acquire the 2D spectrum was about 12and 6 h for the crystalline and glass samples, respectively. The spinningfrequency was stabilized to(10 Hz for all experiments.31P chemicalshifts were referenced relative to 85% H3PO4 solution at 0 ppm.

Results and Discussion

Crystalline Lead Phosphates.To carry out a comprehensiveinterpretation of spectra obtained on glasses, we have firststudied crystalline reference compounds of known structure, forwhich new methods can be validated. Preliminary 1D31P NMRresults have been reported for Pb2P2O7 and Pb(PO3)2.25 We shallfirst report our results obtained for Pb2P2O7, Pb3P4O13, and Pb-(PO3)2 using31P MAS NMR and two-dimensional31P double-quantum NMR.

The high spinning rate31P MAS spectra (10 kHz) of thereference samples Pb2P2O7, Pb3P4O13, and Pb(PO3)2 (not shown)exhibit resolved sharp lines from which isotropic chemical shiftscan be easily measured. These crystalline lead phosphates giveexamples of Q1 and Q2 phosphorus sites with isotropic shiftsin the ranges-8, -12 ppm and-23, -32 ppm, respectively.These values are in agreement with the31P chemical shift trendsdescribed in the literature.3 The chemical shift anisotropy (CSA)tensors of the31P crystalline sites were determined from thespinning sideband intensities in MAS spectra26 with samplespinning at 3 kHz, neglecting the residual dipolar interaction.The experimental slow spinning31P MAS spectra and theirsimulations, according to the parameters given in Table 1, arepresented in Figure 2. It can be noticed that the distinctionbetween Q1 and Q2 resonances can also be made on the basis

(22) Gottwald, J.; Demco, D. E.; Graf, R.; Spiess, H. W.Chem. Phys. Lett.1995, 243, 314-323.

(23) Ernst, R. R.; Bodenhausen, G.; Wokaun, A.Principles of NuclearMagnetic Resonance in One and Two Dimensions; Clarendon Press:Oxford, 1987.

(24) Grandinetti, P. J.RMN, v1.2.1; Department of Chemistry, The OhioState University, Columbus, OH 43210-1173 USA (http://chemis-try.ohio-state.edu/∼grandinetti/RMN).

(25) Prabhakar, S.; Rao, K. J.; Rao, C. N. R.Chem. Phys. Lett.1987, 139,96-102.

(26) Herzfeld, J.; Berger, A. E.J. Chem. Phys.1980, 73, 6021-6030.

Figure 1. (a) Pulse sequence and coherence pathway used for the two-dimensional double-quantum MAS experiments. (b) Detail of the C7sequence used for double-quantum excitation and reconversion. Itinvolves seven phase-shifted pulse cycles timed to span two rotorperiods, each cycle consisting of two (2π) pulses with opposedphases.10

Table 1. Isotropic and Anisotropic31P Chemical Shifts of theCrystalline Lead Phosphates Pb2P2O7, Pb3P4O13, and Pb(PO3)2 (δISO

( 0.5 ppm,Ω ( 2 ppm, K ( 0.02) with Ω ) δ11 - δ33, K ) 3(δ22

- δiso)/Ω36

compoundδISO

(ppm)Ω

(ppm) Kδ11

(ppm)δ22

(ppm)δ33

(ppm)

Pb2P2O7 A -8.4 133.2 -0.69 73.6 -39.2 -59.6B -8.9 130 -0.69 71.1 -38.9 -58.9C -11.3 140.2 -0.38 67.7 -29.1 -72.5D -12 142 -0.38 68 -30 -74

Pb3P4O13 E -9.9 147.3 -0.38 73.1 -28.6 -74.2F -10.8 158.4 -0.33 77.2 -28.4 -81.2G -23.9 209.4 0.38 67.2 2.6 -141.9H -25.3 201.6 0.33 64.3 -2.9 -137.3

Pb(PO3)2 I -27.8 207.7 0.38 62.9 -1.5 -144.8J -28.7 204.1 0.38 60.4 -2.8 -143.7K -30.5 224 0.43 65.5 1.5 -158.5L -31.8 216.5 0.38 62.75 -4.35 -153.8

Order in Crystalline and Glass Lead Phosphates Inorganic Chemistry, Vol. 38, No. 23, 19995213

Page 3: High-Resolution Double-Quantum               31               P MAS NMR Study of the Intermediate-Range Order in Crystalline and Glass Lead Phosphates

of their chemical shift anisotropies (sign of the skew parameter(K) and span of the chemical shift anisotropy).

The structure of the lead pyrophosphate Pb2P2O7 containsfour crystallographically inequivalent Q1 units linked togetherto form two [P2O7]4- anions.27 The 1D MAS NMR spectrumshown in Figure 2 evidences four isotropic peaks, A, B, C, andD at -8.4, -8.9, -11.3, and-12 ppm, respectively, thatcorrespond to the four Q1 sites present in the structure. The 2Ddouble-quantum spectrum of Pb2P2O7 is displayed in Figure 3a.This spectrum exhibits two intense cross-correlation peaks (A-C) and (B-D) that indicate dipolar connectivity between thetwo inequivalent Q1 sites of [P2O7]4- groups. In addition, fourweak autocorrelation peaks appear on the diagonal of the double-quantum spectrum and reflect the spatial proximity of theequivalent [P2O7]4- anions. The intensity ratio between crossand autocorrelation peaks is about 4:1. In the Pb2P2O7 structure,the internuclear P-P distances between chemically linked Q1

units (P-O-P bond) are approximately 0.3 nm and aresignificantly shorter than the other next nearest P-P distancesthat exceed 0.4 nm, since dipolar interaction varies as 1/r3. Thisallows to differentiate dipolar coupling between chemicallylinked PO4 groups and weaker long-range dipolar interactionin the double-quantum spectrum and thus to identify clearlythe P-O-P connectivity scheme.

The structure of the lead tetrapolyphosphate Pb3P4O13 is madeof two inequivalent Q1 units and two inequivalent Q2 unitsassociated together to form a linear [P4O13]6- anion.28 The MASNMR spectrum shown in Figure 2 clearly exhibits four distinctcontributions in agreement with the structure. The two isotropicpeaks at-9.9 and-10.8 ppm, labeled E and F, respectively,are attributed to the two Q1 sites and the two isotropic peaks at-23.9 and-25.3 ppm, labeled G and H, respectively, cor-respond to the Q2 resonances. The double-quantum spectrumof Pb3P4O13 presented in Figure 3b shows three intense cross-correlation peaks E-G, F-H, and G-H, that reflect the dipolarconnectivity between chemically bound PO4 tetrahedra. Theweaker autocorrelation peaks observed on the double-quantumdiagonal correspond to weaker dipolar coupling between nearbybut not linked equivalent Qn groups. This double-quantumspectrum thus evidences a connectivity scheme E-G-H-F andis characteristic of a linear tetrameric [P4O13]6- anion.

The structure of the lead metaphosphate Pb(PO3)2 consistsof infinite phosphate chains containing four inequivalent PO4

tetrahedra.29 Four isotropic resonances at-27.8,-28.7,-30.5,and -31.8 ppm, labeled I, J, K, and L, respectively, aredisplayed in the MAS NMR spectrum (Figure 2) and correspondto these four distinct Q2 sites. The double-quantum spectrumof Pb(PO3)2 is shown in Figure 3c. In this case, the smalldifference between the coupled spin pairs resonances lead toan unresolved spectrum in the double-quantum dimension.However, the correlation peaks can be determined by carefullyanalyzing separated 1D slices of the 2D double-quantumexperiment. Following this analysis, the 2D spectrum indicatesconnectivities between resonances I-K, J-K, I-L, and J-L.In analogy with Pb2P2O7 and Pb3P4O13, the weaker diagonalautocorrelation intensities illustrate the spatial proximity of non-chemically bound PO4 groups. This spectrum is thus charac-teristic of a linear or cyclic phosphate chain with fourinequivalent Q2 sites per period.

From the study of these crystalline compounds and on thebasis of their known structures,27-29 we clearly show that thehigh-resolution double-quantum/single-quantum homonucleardipolar correlation spectroscopy can be used to selectively probethe P-O-P connectivity scheme of crystalline or amorphousphosphate networks.

Lead Phosphate Glasses.As expected, the MAS NMRspectra of31P in the glasses are less resolved than those of thecrystalline samples. Figure 4 shows the evolution of the31P 1DMAS NMR spectra (10 kHz) of the PbO-P2O5 glasses as afunction of the lead content from 67 to 50 mol % PbO. Thesespectra exhibit partly overlapping isotropic resonances with theirassociated spinning sidebands due to an incomplete averagingof the CSA interaction under magic angle spinning. These threedistinct isotropic contributions can be assigned to Q2, Q1, andQ0 units according to the31P chemical shift ranges in crystallinelead phosphates. In agreement with previous works,25 the MASspectrum of the metaphosphate glass is dominated by an intenseisotropic peak at about-24 ppm that correspond to Q2 unitsinvolved in long phosphate chains or cycles. As the PbO contentincreases, the intensity of the Q1 resonance, centered at about-9 ppm, increases indicating a progressive depolymerizationof the phosphate network. The last isotropic component at 1.5ppm is attributed to the Q0 units, in analogy with the31Pisotropic shift of the [PO4]3- group in Pb3(PO4)2 (δISO ) -0.2ppm).

The experimental 1D MAS spectra can be satisfactorilysimulated (including spinning sidebands) under simple hypoth-esis, i.e., assuming Gaussian line shapes for each Qn unit. Forthe metaphosphate glass, a low Q3 intensity at-38 ppm wastaken in account in our simulation. From these simulations, therelative Qn site populations were obtained for each glasscomposition. Table 2 gives the isotropic chemical shifts, linewidths, and relative Qn concentrations for the different samples.Figure 5 shows the evolution of the experimentally determinedQn populations from which we have calculated the equilibriumconstantskn of the disproportionation reactions:30,31

The obtained equilibrium constants arek2 ) 0.001( 0.002andk1 ) 0.022( 0.013. For the glasses having a PbO content

(27) Mullica, D. F.; Perkins, H. O.; Grossie, D. A.; Boatner, L. A.; Sales,B. C. J. Solid State Chem.1986, 62, 371-376.

(28) Averbuch-Pouchot, T.; Durif, A.Acta Crystallogr.1987, C43, 631-632.

(29) Jost, K. H.Acta Crystallogr.1964, 17, 1539-1544.(30) Parks, J. R.; Van Wazer, J. R.J. Am. Ceram. Soc.1957, 79, 4890-

4897.(31) Stebbins, J. F.Nature1987, 330, 465-467.

Figure 2. 31P MAS NMR spectra (3 kHz spinning rate) of Pb2P2O7,Pb3P4O13, and Pb(PO3)2. The asterisks mark isotropic peaks.

2Qn T Qn+1 + Qn-1

5214 Inorganic Chemistry, Vol. 38, No. 23, 1999 Fayon et al.

Page 4: High-Resolution Double-Quantum               31               P MAS NMR Study of the Intermediate-Range Order in Crystalline and Glass Lead Phosphates

close to 50 mol %, the Qn distribution is close to a binarydistribution (k2 ≈ 0) as expected for an ionic modifier cation.30

However, the equilibrium constant calculated for the dispro-portionation reaction of Q1 species indicates a more randomconfiguration of bridging and non bridging oxygen atoms asthe lead content increases. This evolution can be attributed tothe appearance of more covalent Pb-O bonds in the glassstructure, in full agreement with our interpretation of two-

dimensional207Pb PASS NMR spectra that evidence the mixednetwork former/network modifier role of Pb2+ in P2O5-PbOglasses with high lead content.21 It should be noticed that theseequilibrium constants are very similar to those determined inzinc phosphate glasses32 and are 1 order of magnitude weakerthan those measured in lead silicate glasses.33

For each type of Qn unit present in the glass, the line widthis related to a continuous distribution of isotropic chemical shift.This inhomogeneous broadening arises from structural disorder

(32) Brow, R. K.; Tallant, D. R.; Myers, S. T.; Phipher, C. C.J. Non-Cryst. Solids1995, 191, 45-55.

(33) Fayon, F.; Bessada, C.; Massiot, D.; Farnan, I.; Coutures, J. P.J. Non-Cryst. Solids1998, 232-234, 403-408.

Figure 3. High-resolution31P double-quantum MAS spectra of Pb2P2O7 (a), Pb3P4O13 (b), and Pb(PO3)2 (c). The contour levels were set to 14.8,20.8, 29.1, 40.8, 57.1, and 80% of the maximal peak intensity. Asterisks mark impurities.

Figure 4. 31P MAS NMR spectra (10 kHz spinning rate) of the leadphosphate glasses. The asterisks mark isotropic peaks.

Table 2. Isotropic31P Chemical Shifts (δISO), Line Widths (FullWidth at Half-Maximum, fwhm), and Relative Intensities (I%) ofthe Qn Units in the PbO-P2O5 Glasses (δISO ( 0.2 ppm, fwhm(0.5 ppm,I ( 5%)

Q3 Q2 Q1 Q0

mol %PbO δISO (fwhm) I δISO (fwhm) I δISO (fwhm) I δISO (fwhm) I

50 -38.3 (13.2) 3.5-24.6 (9.9) 93.7 -9.7 (7.1) 2.8 -55 - -24.0 (9.8) 74.5 -9.2 (7.3) 25.5-59 - -23.2 (9.8) 54.4 -8.9 (7.2) 44.6 1.5 (7.1) 1.066 - -20.5 (10.0) 13.9-8.3 (7.0) 77.1 1.5 (7.1) 9.067 - -20.1 (10.0) 9.9 -8.0 (7.8) 75.8 1.5 (7.2) 14.3

Order in Crystalline and Glass Lead Phosphates Inorganic Chemistry, Vol. 38, No. 23, 19995215

Page 5: High-Resolution Double-Quantum               31               P MAS NMR Study of the Intermediate-Range Order in Crystalline and Glass Lead Phosphates

such as bond angles, bond lengths variations and highercoordination sphere disorder. It should be noticed that the Qn

line width increases significantly with the number of bridgingoxygen atoms per Qn units (Table 2). As in the case of otherphosphate glasses,34,35 we can also remark that the individualQn isotropic shifts depend on glass composition. It increaseslinearly (31P deshielding) with PbO content, with an increasingslope going from Q0 to Q2.

This description of the glass structure is limited to the short-range order. As shown for the crystalline samples, it is possibleto selectively probe the dipolar couplings between chemicallylinked PO4 tetrahedra using double-quantum NMR experiments.We can then describe the Qn connectivity scheme and theintermediate range order of the glass network.

The 2D double-quantum spectra of the PbO-P2O5 glassesare presented in Figure 6. The double-quantum spectrum of theglass containing 66 mol % PbO (Figure 6a) is composed ofQ1-Q2 cross-correlation peaks and an intense Q1-Q1 autocor-relation peak typical of a [P2O7]4- unit. No Q2-Q2 autocorre-lation peak is evidenced in the spectrum. This indicates thatthe Q2 species are mainly linked to two Q1 units to form a[P3O10]5- group. We also remark that, as expected, the Q0

resonance, which is present in the 1D MAS spectrum, is filteredout of the double-quantum spectrum. In this case, a qualitativeanalysis of both MAS and double-quantum spectra shows thatthe phosphate network is mostly constituted of [P2O7],4-

[PO4]3-, and [P3O10]5- groups. This is in contrast with thePb2P2O7 crystalline structure (of same composition) that containsonly [P2O7]4- anions.

The double-quantum spectrum of the glass containing 59 mol% PbO (Figure 6b) exhibits three separated correlation peaks.

The two intense resonances, located at approximately-31 ppmand -48 ppm in the double-quantum dimension, correspondrespectively to the Q1-Q2 and Q2-Q2 pairs that are involvedin phosphate chains of moderate length. The last significantcontribution (at about-15 ppm) is attributed to the [P2O7]4-

diphosphate groups. This spectrum thus indicates a chain lengthdistribution in the glass network. It is important to remark thesplitting of the Q2-Q2 autocorrelation peak along the diagonalof the spectrum. This splitting indicates that the Q2-Q2 dipolarinteraction involves two structurally different Q2 sites havingdistinct isotropic chemical shifts (the average chemical shiftdifference between two coupled Q2 nuclei is about 2.5 ppm).

As shown in Figure 6c, the Q2-Q2 and Q2-Q1 correlationpeaks are clearly resolved in the double-quantum spectrum ofthe glass containing 55 mol % PbO. In this case, the remainingvery weak Q1-Q1 contribution is attributed to long-range dipolarcouplings.

The double-quantum spectrum of the metaphosphate glass(Figure 6d) exhibits a single intense autocorrelation peakrevealing Q2-Q2 connectivities. The structural motifs consistentwith the spectrum are linear (or cyclic) long phosphate chainsas those involved in the Pb(PO3)2 crystalline structure. Inanalogy with the two previous samples, we observe a splittingof the Q2-Q2 autocorrelation peak along the diagonal.

We mentioned above that the isotropic chemical shift andthe line width of the Qn units depend on glass composition andthus glass network structure. From the double-quantum experi-ments, we also remark that the auto (Qn-Qn) and cross-correlation (Qn-Qm) peaks are located at different isotropicpositions in the single quantum dimension. This suggests thatthe Qn isotropic chemical shift is directly influenced by the typeof the linked adjacent Qn unit. A similar trend has recently beenobserved in the double-quantum spectra of Na2O-P2O5

17 andCaO-P2O5

18 glasses. The Qn species can be thus classifiedaccording to their connectivities, Qn,ij, where the additionalsubscripts describe the type of the bonded PO4 group.18 Asdescribed above, the Qn line widths and chemical shift rangesincrease with the number of bridging oxygen atoms per Qn units.It should be pointed out that, in a similar way, the number ofthe possible local configurations around the Qn species increaseswith its number of bridging oxygen atoms: a Q1 group can bebound to a Q1 (Q1,1) or a Q2 unit (Q1,2), while a Q2 group canform Q2,11, Q2,12, Q2,22, Q2,23, or Q2,33species. The Qn,ij isotropicshifts and line widths, reported in Table 3, were determinedfrom the positions of the auto and cross-correlation peaks inthe single quantum dimension. We observe that changing thenature of the neighboring Qn unit results in an average shift ofabout 2 ppm for both Q1,i and Q2,ij species. The Qn,ij line widthsindicate a strong overlap of their respective chemical shiftranges, in agreement with the results obtained for the crystallinecompounds: the Q1,1 isotropic shifts, measured for Pb2P2O7, liesbetween-8.4 and -12 ppm while the Q1,2 sites found inPb3P4O13 lead to resonances at-9.9 and-10.8 ppm.

Since the excitation of double-quantum coherence can hardlybe quantitative, the determination of the relative Qn,ij concentra-

(34) Brow, R. K.; Kirkpatrick, R. J.; Turner, G. L.J. Non-Cryst. Solids1990, 116, 39-45.

(35) Losso, P.; Schnabel, B.; Ja¨ger, C.; Sternberg, U.; Stachel, D.; Smith,D. O. J. Non-Cryst. Solids1992, 143, 265-273.

(36) Mason, J.Solid State NMR1993, 2, 285-288.

Table 3. Isotropic31P Chemical Shifts (δISO), Line Widths (Full Width at Half-Maximum, fwhm), and Relative Intensities (I ( 7%) of the Qn,ij

Units in the PbO-P2O5 Glasses

Qn,ij unit Q0 Q1,1 Q1,2 Q2,11 Q2,21 Q2,22

δISO (ppm) 1.5( 0.5 -8.0( 0.5 -9.5( 0.5 -20.0( 0.5 -22.5( 0.5 -24.5( 0.5fwhm (ppm) 7.0( 0.5 7.0( 0.5 7.0( 0.5 8.5( 0.5 8.5( 0.5 9.0( 0.555 mol % PbO, intensity (%) - 1.5 23.7 3.0 17.7 54.159 mol % PbO, intensity (%) 0.9 13.6 30.7 6.0 18.7 30.166 mol % PbO, intensity (%) 8.7 59.3 19.2 7.9 3.4 1.567 mol % PbO, intensity (%) 14.6 58.7 17.1 7.5 2.1 -

Figure 5. Experimentally determined Qn distribution in lead phosphateglasses as a function of the [PbO]/[P2O5] ratio. The fitted lines werecalculated from the equilibrium constantsk2 ) 0.001 andk1 ) 0.022.

5216 Inorganic Chemistry, Vol. 38, No. 23, 1999 Fayon et al.

Page 6: High-Resolution Double-Quantum               31               P MAS NMR Study of the Intermediate-Range Order in Crystalline and Glass Lead Phosphates

tions from double-quantum spectra is not straightforward.Nevertheless, it is possible to determine these concentrationsfrom the simulations of the quantitative 1D MAS spectra usingthe parameters obtained from 2D spectra and constraining theQn,ij positions and line widths to vary within a range of(0.5ppm. Moreover, the stoechiometry of the system imposesanother constraint that is [Q1,2] ) 2[Q2,11] + [Q2,21]. Followingthis protocol, we have estimated the relative populations of Qn,ij

units given in Table 3. These relative Qn,ij populations are ingood agreement with the previous quantification of Qn species.For PbO content higher than 55 mol %, the phosphate networkis composed of Q2, Q1, and Q0 species. For this compositionalrange, only the Q2 and Q1 species have P-O bridging bonds.Thus, assuming a homogeneous distribution of Q2 and Q1 unitsin the glass, a random phosphorus connectivity scheme can bedescribed using a binomial distribution of Q2 and Q1 bridgingbonds as proposed by Alam and Brow.16 As shown in Figure7, the experimentally determined Qn,ij populations are consistentwith those calculated from the Qn populations assuming abinomial distribution of connectivities in the glass structure. Itshould also be pointed out that for the glasses with high leadcontent, the phosphate chain-length distribution can be deducedfrom these Qn,ij concentrations, as discussed by Witter et al..18

These results show that the high-resolution double-quantumNMR experiment can be used to improve the structuraldescription of the phosphate network and to obtain intermediaterange order information in phosphate based glasses.

Figure 6. 31P double-quantum MAS spectra of the glasses 0.66PbO-0.34P2O5 (a), 0.59PbO-0.41P2O5 (b), 0.55PbO-0.45P2O5 (c), and 0.50PbO-0.50P2O5 (d). The contour levels were set to 16.7, 23.4, 32.8, 45.9, 64.2, and 90% of the maximal peak intensity.

Figure 7. Experimentally determined Qn,ij distributions in leadphosphate glasses as a function of the [PbO]/[P2O5] ratio. The fittedlines were calculated from the Qn populations assuming a binomialdistribution of connectivities in the phosphate network.

Order in Crystalline and Glass Lead Phosphates Inorganic Chemistry, Vol. 38, No. 23, 19995217

Page 7: High-Resolution Double-Quantum               31               P MAS NMR Study of the Intermediate-Range Order in Crystalline and Glass Lead Phosphates

Conclusion

Using solid-state31P MAS and 2D double-quantum NMR,we have investigated the local and intermediate range orderingaround phosphorus in binary PbO-P2O5 glasses. The distribu-tion of the Qn species, determined from the simulation of 1DMAS spectra, indicates a significant disproportionation reaction(2Qn T Qn+1 + Qn-1) near the pyrophosphate composition,which is attributed to the appearance of more covalent Pb-Obonds in the glass structure. As shown in the case of crystallinereference samples, the connectivities between Qn groups canbe determined from 2D double-quantum spectra. This allowedus to probe the phosphorus connectivity scheme in the glass

network and to show that the Qn chemical shift is directlyinfluenced by the type of the adjacent PO4 tetrahdron (Qn,ij

species). By simulating the quantitative 1D MAS spectra usingQn,ij chemical shifts obtained from double-quantum spectra, therelative Qn,ij concentrations and their evolutions with the glasscomposition have been estimated.

Acknowledgment. This work was supported by CNRS (UPR4212) and Re´gion Centre. The authors thank C. Ja¨ger for usefuldiscussion.

IC990375P

5218 Inorganic Chemistry, Vol. 38, No. 23, 1999 Fayon et al.